For AI agents: a documentation index is available at /llms.txt. Append .md to any page URL for markdown, or send Accept: text/markdown.
Experiment Groups and Parameters
Define a Statsig experiment's variant groups and the parameter values your code reads, and understand which parts of the group structure you can change after the experiment starts.
Groups are the variant labels in the console, such as Control or Test. Parameters are the values your code reads to change the product or feature's behavior. Configure your parameters before your groups on the experiment's Setup tab. For why your code should read parameters rather than check group names, go to Getting the group.
Add groups and set parameter values
Enter the parameter values for each variant. To add a group, click the + icon next to the existing groups. Statsig adjusts the user allocation automatically as you add groups.
In the Groups section, you can also name, describe, and add variant images for each group. Group names and descriptions are labels that help you identify and compare variants in the Statsig console. Your code reads parameter values through the SDK and never reads group names, so you can rename or re-describe a group at any time without a code change or any effect on what users experience.
What you can change after the experiment starts
Set the group structure before you start the experiment. After an experiment starts, you can't add groups, delete groups, or change their percentage splits.
| Change | After the experiment starts |
|---|---|
| Add or delete a group | Not available |
| Change a group's percentage split | Not available |
| Edit a group's parameter values | Available |
| Rename or re-describe a group | Available |
| Stop one group from serving its values | Available through Disable a Group |
To change the group structure of a running experiment, reset it or create a new experiment. Both re-randomize assignment and end the current analysis window.
Editing parameter values during a run can mix materially different treatments into one scorecard. Parameters define the values the SDK returns; they aren't a result dimension or a version label. If the treatment's meaning changed, start a new iteration instead.
Was this helpful?